New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
Have you seen their hardware requirements? It's literally millions in hardware leases, no ones going to take that kind of risk unless they might have a few clients that could use it. The only customers that might be interested in 768GB RAM nodes is SAP HANA people.
Francisco
I meannnnnnnn they are using small instances on AWS most likely somebody could slap some common sense into them, Parler could run on maybe 5 to 10 good servers. Probably this is the result of some hipster programming preference.
I stand corrected , after checking the requirements again I noticed I read the node RAM as MB instead of GB. I mean wasn't parler kind of small? Why so much?
Wow. Must be some amazingly inefficient code!
I'm assuming they don't know what classifies as a 'good index' in their database, and literally just have every field indexed, causing them to require wild amounts of RAM for their key buffers.
Given the scraping people did, they have 70 - 80TB of data. If you tally the amount of RAM those database servers have... it comes out right around the same.
Francisco
The cloud will set you free... From optimization.
JFC 😳
Get back to writing PWAs. Apple still support that?
And likely a lot of CPU to update all those indexes on every INSERT/UPDATE.
I'm sure AWS isn't slapping 768GB RAM on top of some E5 v2's or something :P
Still, the hardware list shows that they don't have a proper DBA. That, or their developers are doing some hellishly brutal filtering (select 100,000 rows from a table and then if() through it trying to find 3 rows) to require so many CPU's.
Francisco
Society of the day.
Do you really think 76 million people got booted from twitter?
Not every Republican is on board with the overt white nationalism and QAnon - those people didn't get banned from Twitter, and weren't on Parler
Parler had 2,000,000 users, 0.1% is 2000 users - they got way more than that in QAnon/Proud Boy/etc signups after Twitter booted them. The site has long been a haven for the far-right because it has tolerated violent rhetoric.
In the free market, businesses can choose who they do business with. Amazon is free to decide that it doesn't want to host far-right platforms - they tend to be a liability, particularly if people are freely using the platform to organize more violence after a first terror attack on Congress.
Likewise, Twitter doesn't want to deal with repercussions of hate groups organizing on their platform
No, it's more about economics - hosting terrorists on your platform is very bad for PR, which is bad for business - this isn't even counting the possibility of lawsuits Amazon could face
Yeah, I've heard this argument before.
I mean, this is a site where every provider has rules about what they will and will not tolerate on their platform. All of us has at one point or another defended another company here who has booted a client that caused problems or brought unwanted attention. Yet, every time a far-right group goes too far and causes trouble, there is always a contingent of users who suddenly set a double standard of "but but but free speech!", as if because their views are political in nature, they are entitled to host with whatever provider they want.
Really, it should always be understood that when you cause trouble, you get treated like a troublemaker. All that is guaranteed by the 1st amendment is freedom from government reprisal for expressing your beliefs, provided they do not infringe on the rights of others. It does not somehow provide you freedom from consequences.
So, what is a "far-right" group? What is a terrorist?
Does "one guy in a group (allegedly) killed a cop!" show that that group and everyone who looks like them (e.g red MAGA cap, flag) are terrorists and violent?
Are you doing this "for the lulz"? You perfectly know where is limit between free speech and extremism. Both right and left.
Mods please dump this thread to Cest Pit as per rules of LET.
@LTniger
I asked him, well noted after multiple comments begging to have a closer look.
FWIW: Here's the relevant bit from an official DDoS-GUARD statement that stated that their business is DDOS-protection and not hosting.
Lowendtalk and WHT are town squares of low-end hosting businesses. Let’s be real, if you are banned LET/WHT and customers leaved bunch of negative reviews, you are dead.
I wonder how long Parler would stay up if this documentary was in trending
@vimalware
As you seem to not have read or understood it ...
Ouch, I felt that! Its not even an exaggeration... :-/
The American government supports destruction and violence against other countries, as long it serves themselves.
In the wonderland USA the situation is even worse. They only support people like Navalnyi cause they create discord in the east. As always their only target is to have destructive activity around the world rather than developement so they can stand proud above others even without ever evolving as a nation.
I'm not saying Russia is so awesome. But it just looks biased when fingers are not pointed consistently. I also always find it self-righteous when one party is allowed to get away with all sorts of mischief, while expecting the other party to just turn the other cheek. If you punch me 10 times and then I punch you back, I'm not the bad guy.
As for left vs. right and all that, I'm currently listening to this (non-political) podcast, and he puts it better than I could: from 2:01:00
Book recomendation: Manufacturing Consent: The Political Economy of the Mass Media by Noam Chomsky and Edward Herman
The US is known to support dictators in the name of stability, not anarchy and destruction.
What the fuck are you talking about? Are you saying US punches Russia 10 times and Russia punches back once? It's unclear wtf your point is, but if that was, you're being silly.
My recommendation is to watch the entire series "The Americans" to better understand US vs Russia (if not for Keri Russell's perfect ass).
If their DB was indexed properly is this still even doable without using some sort of elastic cloud provider like AWS?
The database is forever expanding and at a pretty remarkable rate. Now you have to replicate it, back it up, forever tweak the settings, etc.
Won't they be ** always ** running out of disk space, dealing with hardware failures, praying whatever failover they cobbled together works, etc?
I dk.. from my experience with MYSQL and hundreds of thousands of INSERTS a day, it becomes an engineering challenge to keep the DB on the hardware.
Is it possible to just rent massive dedicated servers at 8-10 providers across the globe to replicate the databases and just have the front end at one of these sketchy bulletproof hosters? What does failover, internet bottlenecks, and SELECT speeds look like when the front end of the site and the DB servers are scattered over the planet? i.e. the front end is in some random "bulletproof" host somewhere in eastern Europe and your DBs are all with providers in the US?
Seems like a harder engineering challenge than most here give it credit for.
The layout, colors, and interface on their website were horrible IMO. I'd hate to see their back end scripts based on what I saw of the front end.
If the site comes back with old data in place and continues to work I'd be shocked
Replicating isn't going to work well for postgres/mysql as you'll have the latency of it.
Still, Storage isn't the issue here, this is purely memory. Each node also had 20TB or so of NVME storage, on top of 768GB RAM/each.
Francisco
As I've said, Parler seems to have been technically incompetent and committing the cardinal sin of "thinking" that throwing loads of money at a problem can somehow replace proper design and engineering.
If that kind of a service needs more than 256 GB of RAM and 20 TB (!) of NVMe per node something is awfully wrong.
You don't use MySQL in this type of environment. You use a NoSQL database with relaxed consistency because consistency isn't absolutely critical. Makes it trivial to scale horizontally.
If you're processing credit card transactions, then your DB can never give different answers to two different people, even if they're a millisecond apart, because then the accounting will blow up. If you must have 100% consistency (and there are plenty of apps that need this), then you use a huge server, or scale with something like Oracle RAC, or you use a mainframe, iSeries, etc.
But for a lot of apps, you don't need 100% consistency. If you and I are on other sides of the world and we look at a post at the same time and I see 76 likes and you see 78, who cares? The DB will eventually become consistent.
Big social media environments use a mix of relational databases (MySQL, Pg, whatever) and NoSQL.
Whatever they are doing it's not too trivial because they still aren't back up.
Don't you still need big hardware requirements to scale NoSQL across different datacenters?
I'd be surprised if they ever came back online. It's either too hard to figure it out without big tech or they aren't smart enough, don't have enough $$, or something.
I'm surprised Oracle didn't take them and say, "who the fuck did your database? We can do better, just keep the wheelbarrows of cash coming".
It is now really back online. I mean, it seems to provide a login page.
Reuters sent the news
Who now hosts them? Just curious.